ABSTRACT
A comprehensive life-cycle performance assessment of structures and infrastructures requires the definition of Performance Indicators (PIs), which provide an easier and time efficient assessment procedure. It is a tool by which the safety and integrity of the structural system is ensured by preventing the PIs from crossing their thresholds. While structures are subjected to time-dependent degradation processes which require consideration of uncertainties, the evolution of PIs can be simulated and forecasted with appropriate statistical model in combination with available inspection and monitoring information. An innovative Bayesian framework for making use of the available historical data and incorporation of additional information from inspection and monitoring in a so-called Stepwise Changing Rate Updating Method (SCRUM) is developed in this paper to predict the future PIs. Specific to the approach is the use of the Changing Rate (CR) between subsequent PI observations. First, an estimation of the statistical properties of the PI as well as its CR is made, and then further inspection and monitoring allows for incorporating additional information in the Bayesian updating framework for performance prediction. A time-dependent weight factor is defined and implemented in order to account for the decreasing influence of previous CRs on the predicted PI. The advantage of SCRUM is that neither an underlying mechanical nor dynamic model is needed to do the forecasting, making it an easy applicable tool which only uses Monte Carlo simulations to fulfill its task. The primary objective is to determine the lifetime of the structure, given a clear definition of the lifetime and the failure probability limit. Furthermore, the developed method can be used for the detection of optimal inspection intervals. The application of SCRUM is limited to stable processes, i.e. processes for which the performance indicator decreases almost monotonic.
